Statistics from Altmetric.com
If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.
The paper by Lynch et al 1 is intended to cast doubt on the importance to health of social capital, relative income and psychosocial pathways, and to gain support for a view of the continued importance of absolute income and of “neo-material” factors.
Before answering specific points I should draw attention to two new studies that support the importance of relative income over absolute income in richer countries. Firstly, data for 21 regions of Taiwan showed that, as living standards rose during the course of rapid economic growth, income inequality replaced absolute median income as the best predictor of mortality.2 Secondly, in an analysis of infant mortality in relation to income distribution and gross national product per capita (GNPpc) in three sets of data covering richer and poorer countries, Hales et al 3 showed that income distribution was more important than GNPpc except in the poorer countries.
Relative and absolute income
Lynch et al show a positive correlation between life expectancy and gross domestic product per capita (GDPpc) in member countries of the Organisation for Economic Cooperation and Development (OECD) where I had shown none.4 Their evidence misses the point. The difference arises because OECD has added new and poorer members. As I have emphasised, absolute living standards are important in poorer countries. Thus, if we take combined male and female life expectancy in relation to GDPpc at purchasing power parities among the richest 21 countries in 1995,5 instead of finding the positive correlation reported by Lynchet al, there is a weak negative one (r= −0.11). Alternatively, taking just the 23 countries with the highest life expectancy (74 years or above), the correlation between GDPpc and life expectancy is also non-existent (r=0.08). The positive correlation reported by Lynch et al therefore results wholly from the inclusion of …